Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

AI success requires leaders who practice transformative change

The internet, search, mobile, and social media taught us everything we need to know about implementing transformative technology.

[Photo: Getty Images]

Profile photo of maria-fernanda-cala-luque

A recent MIT/NANDA report claiming that 95% of AI projects fail missed the real culprit: We’re ignoring decades of proven transformative change wisdom.

I’ve seen this movie before. During my time leading the digital transformation of Citi Cards we faced identical challenges implementing what seemed like obviously beneficial technology. The pattern repeats with depressing regularity: internet adoption, search integration, mobile transformation, social media adoption. Each time, organizations stumble not because the technology fails, but because leaders discard what we’ve learned about how transformative change really happens with “this time is different” thinking.

DON’T SKIP THE EXPLORATION PHASE

The MIT report attributes AI “failure” to technical limitations—what they call a “learning gap.” But the report missed the human gap entirely. Every major technology wave begins with an exploration phase that requires discipline, patience, and tolerance for apparent failure. Instead, organizations jump straight to ROI demands and permanent solutions.

This was our experience transforming Citi Cards. Success came from building coalitions of passionate supporters who could experiment through corporate resistance. We focused on identifying customer behaviors that drive key business metrics—completion of applications to get a card, engaging in steps towards repaying a past-due loan, and many others—then tested how intentionally designed user experiences enabled by technology could influence those behaviors. One-and-done pilots as the benchmark of success is the wrong approach entirely.

THE “SHADOW ECONOMY” PREDICTS SUCCESS

Something fascinating emerges in every technology wave: Employees adopt tools personally before organizations catch up. The MIT researchers discovered what they call a “shadow AI economy”—employees using personal ChatGPT accounts, Claude subscriptions, and other consumer tools to handle significant portions of their jobs, often without the IT department’s knowledge or approval.

This shadow usage isn’t a problem to solve—it’s early intelligence about what actually works. The organizations that recognize and build on personal adoption patterns position themselves ahead of competitors still debating AI strategy.

This has happened before. Mobile applications were dragged into workplaces by employees’ personal usage. Search transformed how people accessed information before organizations figured out enterprise search strategies. Social media revolutionized customer service because customers demanded it, not because companies anticipated its value.

RIGID ROI CAN KILL INNOVATION

The MIT researchers focused on model quality and technical integration while dismissing human factors entirely. This represents a fundamental misunderstanding of how transformative change happens. Technology adoption follows predictable patterns that we’ve documented for three decades.

First, passionate early adopters experiment and build credibility through visible results. They establish coalitions of supporters who share the passion and drive to work through organizational barriers. These champions generate small wins around use cases that matter to decision makers, building trust incrementally rather than demanding massive resource commitments upfront.

The wrong approach—which the MIT report inadvertently validates—imposes rigid ROI frameworks before understanding what behaviors drive business value. This kills innovation before it can demonstrate impact.

LEARN FROM THREE DECADES OF DIGITAL TRANSFORMATION

Each technology wave teaches the same lessons, yet we keep ignoring institutional knowledge. The winners were organizations that connected what technology made possible with what people would actually embrace. They hired for critical thinking and creativity rather than technical proficiency, recognizing that today’s tools become tomorrow’s legacy systems.

Most importantly, they overcame the inclination to impose ROI metrics too early. Instead, they identified the customer, employee, and stakeholder behaviors that drive the metrics reported in financial statements, then tested how technology influenced those behaviors through careful experimentation.

THE REAL FRAMEWORK FOR AI SUCCESS

Based on three decades of technology waves, successful AI implementation requires abandoning the “big bang” mentality in favor of disciplined experimentation.

Start with cross-functional teams that bring different knowledge and experiences to identifying opportunities. Set ambitious but not “suicidal” aspirations that push thinking without creating impossible standards. Focus intensively on the human factors—employee fears, workflow integration, and trust-building—that the MIT report ignored entirely.

Address the very real human concerns about displacement and relevance rather than dismissing them as resistance to be overcome.

FINAL THOUGHTS

We can continue repeating the same mistakes or we can apply institutional knowledge that already exists. The internet, search, mobile, and social media taught us everything we need to know about implementing transformative technology. AI isn’t different—it’s just the latest test of whether we’ve learned anything from our own history.

The path to AI impact isn’t through better technology. It’s through better change leadership, applied with the wisdom of experience rather than the arrogance of “this time is different.”

Amy Radin is the founder of Pragmatic Innovation Partners LLC, a keynote speaker, and Columbia University instructor.